The G protein-coupled receptors (GPCRs) are the largest family in the mammalian genome, and are critical to a number of cell signaling processes. As a result, they are of enormous biomedical importance; by some estimates, as many as 50% of new pharmaceuticals target GPCRs. Unsurprisingly, there has been a huge research investment in understanding their biophysics. However, integral membrane proteins are challenging to work with experimentally, leaving an opportunity for computational methods to make a significant contribution. We will use multiscale modeling techniques, including all-atom molecular dynamics simulations and elastic network models, to explore the behavior of several GPCRs, including rhodopsin (and its retinal-free form, opsin) and the 2-adrenergic receptor (B2AR). Specifically, we will investigate the role of ligand binding in modulating GPCR function, via two separate all-atom molecular dynamics calculations. Microsecond-scale simulations of opsin will, when contrasted with our previous work on rhodopsin in the dark state and during the early stages of activation, allow us to see which interactions in rhodopsin are determined by the presence of the ligand, while the planned simulations of the full activation process will give the first atomic-level view of the structural changes involved in GPCR activation; this knowledge could be critical to the design of novel inhibitors to other GPCRs. The second goal of this proposal is to clarify the role of internal waters in the activation mechanism of GPCRs; our previous simulations described significant increases in the internal hydration of rhodopsin and B2AR. Here, we propose to pursue those observations more rigorously, using automatic pattern recognition methods to correlate hydration changes with functionally interesting protein motions in simulations of rhodopsin, B2AR, and the cannabinoid-2 receptor (CB2). The third goal of the proposal is to develop elastic network models - a simple, computationally inexpensive approach where the protein's interactions are represented as a network of springs - in order to explore larger scale problems not readily amenable to all-atom molecular dynamics, like the modulation of protein motions by G protein binding and GPCR oligomerization. A number of possible network model implementations will be considered, and the models will be carefully validated by quantitative comparison to extensive molecular dynamics simulations, including those proposed for the first aim. The fourth and final goal of the proposal is to assess the validity of a common assumption, that rhodopsin is a good template for understanding GPCR activation in general. To test this hypothesis we will apply multiple computational methods, including long timescale molecular dynamics and elastic network models, to a series of GPCRs, including rhodopsin, opsin, B2AR, and CB2. We will quantitatively correlate the fluctuations of the different GPCRs, with the hypothesis that motions conserved across multiple GPCRs are likely to be functionally significant.